Heavy metal in radiation therapy: A simple method to differentiate hip prosthesis materials

Abstract

BackgroundIn recent years, the number of hip replacement patients receiving radiation therapy has steadily increased. In parallel, strategies have been developed to reduce metal artifacts in computed tomography (CT) images and improve the accuracy of dose calculation algorithms. However, in certain situations, knowledge of the type of prosthesis material is required to accurately determine the dose distribution.PurposeThis study aims to identify physical materials in hip prostheses to correctly assign them in the treatment planning system and improve dose calculation accuracy.MethodsWe first verified the validity of the extended CT mass density calibration curve measured on titanium (Ti) and stainless steel (SS) metal inserts of two different diameters. Then using dedicated reference objects of various circular diameters, we developed a method based on interpolation functions to differentiate between Ti and SS material groups. Forty data sets from 18 patients were used to validate our method on two different reconstruction kernels: a standard Br44f and the electron DirectDensity (Sd40f) kernels from Siemens.ResultsHounsfield units (HU) of Ti and SS inserts were found to vary widely depending on insert diameter, CT spectrum, and reconstruction kernels due to cupping artifacts. The largest HU difference (‐79%) was obtained for SS at 70 kV with Br44f when the diameter increased from 8 to 30 mm. Therefore, under these conditions, the extended CT‐density calibration curve is not recommended for heavy metal density determination. Using our interpolation‐based method, we achieved excellent detection (100%) and material differentiation (100%) results for stems in both reconstruction kernels. At CT energies between 110 and 140 kV, the detection and material differentiation rates were 93.3% and 92.9% for the heads and 93.3% and 92.9% for the acetabular cups, respectively, with the Br44f. Similarly, the use of Sd40f resulted in detection and differentiation rates of 94.7% and 100% for the heads and 100% and 95.0% for the acetabular cups, respectively.ConclusionThis method makes it possible to differentiate between hip prosthesis materials and correctly assign them to the Ti or SS group without prior knowledge of the prosthesis type, regardless of the reconstruction kernels. In combination with the Acuros XB (Varian) or Monte Carlo dose algorithms, excellent dosimetric accuracy can be achieved even in the vicinity of hip prostheses. By performing basic measurements, the method can be adapted to other CT units and reconstruction kernels, replacing the use of an extended CT‐density calibration curve.